The Society of General Internal Medicine (SGIM) has released a position statement outlining recommendations for generative AI use in healthcare, emphasizing collaboration between clinicians and technologists while cautioning against rapid workforce displacement.

The statement provides specific guidance for three key stakeholder groups. For clinicians, SGIM recommends maintaining receptiveness to AI tools while applying the same critical evaluation standards used for other medical advancements. The society emphasizes that clinicians must recognize errors and omissions as major technical weaknesses when incorporating these tools into practice.

For technologists, SGIM calls for a significant shift away from the expectation that clinicians will “supervise” AI systems. Instead, developers should hold their technologies to the same high standards expected of the clinical workforce. “These organizations and individuals should strive to design high-performing, well-studied tools that improve care and foster the therapeutic relationship, not simply those that improve efficiency or market share,” the Society wrote.

Healthcare organizations are advised to pursue both incremental and transformative changes while avoiding the rush to replace human workers with AI tools. SGIM recommends focusing AI implementation on preventative care and chronic condition management, areas with significant impact on modern internal medicine practice.

The position statement strongly affirms that medicine remains “a fundamentally human endeavor which should be enhanced by technology, not displaced by it.” It emphasizes the importance of ongoing partnerships between technologists, clinicians, and patients to understand real-world needs and ensure AI systems provide reliable, verifiable outputs.

These recommendations come as regulatory bodies, including the FDA, are developing frameworks for AI oversight in healthcare, highlighting the growing need for clear guidelines on responsible AI implementation in clinical settings.

Show CommentsClose Comments

Leave a comment